111 lines
3 KiB
Ruby
Executable file
111 lines
3 KiB
Ruby
Executable file
#!/usr/bin/env ruby
|
|
|
|
##
|
|
# This script requests each verse, in each chapter
|
|
# of The Qur'an - in the Portuguese language. The content
|
|
# is requested from the https://al-quran.cc website.
|
|
#
|
|
# Each chapter is then saved in a JSON file, for example:
|
|
# "src/pt/<chapter_num>.json"
|
|
|
|
##
|
|
# Dependencies
|
|
require "net/http"
|
|
require "nokogiri"
|
|
require "json"
|
|
require "paint"
|
|
|
|
##
|
|
# Chapter names
|
|
ch_names = %w[
|
|
al-fatiha al-baqara aal-e-imran an-nisa
|
|
al-maeda al-anaam al-araf al-anfal
|
|
at-tawba yunus hud yusuf
|
|
ar-rad ibrahim al-hijr an-nahl
|
|
al-isra al-kahf maryam ta-ha
|
|
al-anbiya al-hajj al-mumenoon an-nur
|
|
al-furqan ash-shuara an-naml al-qasas
|
|
al-ankabut ar-rum luqman as-sajda
|
|
al-ahzab saba fatir ya-seen
|
|
as-saaffat sad az-zamar ghafir
|
|
fussilat ash-shura az-zukhruf ad-dukhan
|
|
al-jathiya al-ahqaf muhammad al-fath
|
|
al-hujraat qaf adh-dhariyat at-tur
|
|
an-najm al-qamar ar-rahman al-waqia
|
|
al-hadid al-mujadala al-hashr al-mumtahana
|
|
as-saff al-jumua al-munafiqoon at-taghabun
|
|
at-talaq at-tahrim al-mulk al-qalam
|
|
al-haaqqa al-maarij nooh al-jinn
|
|
al-muzzammil al-muddaththir al-qiyama al-insan
|
|
al-mursalat an-naba an-naziat abasa
|
|
at-takwir al-infitar al-mutaffifin al-inshiqaq
|
|
al-burooj at-tariq al-ala al-ghashiya
|
|
al-fajr al-balad ash-shams al-lail
|
|
ad-dhuha al-sharh at-tin al-alaq
|
|
al-qadr al-bayyina az-zalzala al-adiyat
|
|
al-qaria at-takathur al-asr al-humaza
|
|
al-fil quraish al-maun al-kauther
|
|
al-kafiroon an-nasr al-masadd al-ikhlas
|
|
al-falaq an-nas
|
|
]
|
|
|
|
##
|
|
# Configuration variables.
|
|
base_uri = "al-quran.cc"
|
|
path = "/quran-translation/portuguese/%{ch_name}/%{verse_num}.html"
|
|
cool_off = 5
|
|
src_path = File.join(
|
|
__dir__, "..", "src", "json", "ar", "%{chapter_num}.json"
|
|
)
|
|
dest_path = File.join(
|
|
__dir__, "..", "src", "json", "pt", "%{chapter_num}.json"
|
|
)
|
|
|
|
##
|
|
# Share a single Net::HTTP instance.
|
|
http = Net::HTTP.new(base_uri, 443)
|
|
http.use_ssl = true
|
|
|
|
##
|
|
# Helper method.
|
|
def get_request(path)
|
|
Net::HTTP::Get.new(
|
|
path,
|
|
"Accept" => "text/html"
|
|
)
|
|
end
|
|
|
|
##
|
|
# Helper method.
|
|
def extract_verse!(res, remove_suratu)
|
|
doc = Nokogiri::HTML(res.body)
|
|
verse = doc.css(".ayah .translate").first
|
|
(remove_suratu ? verse.text.sub(/^Suratu [\w\-.]+/, '') : verse.text).strip
|
|
end
|
|
|
|
##
|
|
# main()
|
|
1.upto(114) do |chapter_num|
|
|
verses = JSON.parse File.read(format(src_path, chapter_num:))
|
|
rows = []
|
|
ch_name = ch_names[chapter_num - 1]
|
|
final_dest = format(dest_path, chapter_num:)
|
|
|
|
print "Fetch: ", ch_name, "\n"
|
|
verses.each.with_index(1) do |_, verse_num|
|
|
htm_file = format(path, ch_name:, verse_num:)
|
|
case res = http.request(get_request(htm_file))
|
|
when Net::HTTPOK
|
|
rows.push([verse_num, extract_verse!(res, verse_num == 1)])
|
|
else
|
|
print Paint["ERROR (#{res.class}): ", :red, :bold], final_dest, "\n"
|
|
break
|
|
end
|
|
end
|
|
File.write(final_dest, JSON.pretty_generate(rows))
|
|
print Paint["OK: ", :green, :bold], final_dest.sub(ENV['HOME'], ''), "\n"
|
|
sleep cool_off
|
|
print Paint["Chill for #{cool_off} seconds", :blue, :bold], "\n", "\n"
|
|
end
|
|
|
|
|